Details | Last modification | View Log | RSS feed
| Rev | Author | Line No. | Line |
|---|---|---|---|
| 716 | lvd | 1 | // http://scale2x.sourceforge.net/algorithm.html |
| 2 | |||
| 3 | #include "std.h" |
||
| 4 | |||
| 5 | #include "emul.h" |
||
| 6 | #include "vars.h" |
||
| 7 | #include "draw.h" |
||
| 8 | #include "dxrend.h" |
||
| 9 | #include "dxrcopy.h" |
||
| 10 | #include "dxrframe.h" |
||
| 11 | #include "dxr_advm.h" |
||
| 12 | |||
| 13 | inline void line_8_any(unsigned char *dst, unsigned char *src) |
||
| 14 | { |
||
| 15 | if (conf.noflic) line8_nf(dst, src, t.sctab8[0]); |
||
| 16 | else line8(dst, src, t.sctab8[0]); |
||
| 17 | } |
||
| 18 | |||
| 19 | inline void line_32_any(unsigned char *dst, unsigned char *src) |
||
| 20 | { |
||
| 21 | if (conf.noflic) line32_nf(dst, src, t.sctab32[0]); |
||
| 22 | else line32(dst, src, t.sctab32[0]); |
||
| 23 | } |
||
| 24 | |||
| 25 | #if 1 // switch between vectorized and branched code |
||
| 26 | |||
| 27 | #ifdef MOD_SSE2 |
||
| 28 | |||
| 29 | void lines_scale2(const unsigned char *src, unsigned y, unsigned char *dst1, unsigned char *dst2, unsigned nPix) |
||
| 30 | { |
||
| 31 | const unsigned char |
||
| 32 | *u = src + ((y-1) & 7)*sc2lines_width, |
||
| 33 | *m = src + ((y+0) & 7)*sc2lines_width, |
||
| 34 | *l = src + ((y+1) & 7)*sc2lines_width; |
||
| 35 | |||
| 36 | for (unsigned i = 0; i < nPix; i += 8) { |
||
| 37 | |||
| 38 | __m64 uu = *(__m64*)(u+i); |
||
| 39 | __m64 ll = *(__m64*)(l+i); |
||
| 40 | __m64 cmp = _mm_cmpeq_pi8(uu,ll); |
||
| 41 | |||
| 42 | if (_mm_movemask_pi8(cmp) != 0xFF) { |
||
| 43 | |||
| 44 | __m128i mm = _mm_loadu_si128((__m128i*)(m+i-4)); |
||
| 45 | __m128i uu = _mm_loadu_si128((__m128i*)(u+i-4)); |
||
| 46 | __m128i ll = _mm_loadu_si128((__m128i*)(l+i-4)); |
||
| 47 | |||
| 48 | __m128i md = _mm_slli_si128(mm,1); |
||
| 49 | __m128i mf = _mm_srli_si128(mm,1); |
||
| 50 | __m128i maskall = _mm_or_si128(_mm_cmpeq_epi8(md,mf), _mm_cmpeq_epi8(uu,ll)); |
||
| 51 | |||
| 52 | __m128i e0, e1, v1, v2, v3; |
||
| 53 | |||
| 54 | e0 = _mm_cmpeq_epi8(md,uu); |
||
| 55 | e0 = _mm_andnot_si128(maskall, e0); |
||
| 56 | e0 = _mm_srli_si128(e0,4); |
||
| 57 | e0 = _mm_unpacklo_epi8(e0, _mm_setzero_si128()); |
||
| 58 | |||
| 59 | e1 = _mm_cmpeq_epi8(mf,uu); |
||
| 60 | e1 = _mm_andnot_si128(maskall, e1); |
||
| 61 | e1 = _mm_srli_si128(e1,4); |
||
| 62 | e1 = _mm_unpacklo_epi8(_mm_setzero_si128(), e1); |
||
| 63 | |||
| 64 | e0 = _mm_or_si128(e0, e1); |
||
| 65 | |||
| 66 | v1 = _mm_srli_si128(mm,4); |
||
| 67 | v1 = _mm_unpacklo_epi8(v1,v1); |
||
| 68 | v2 = _mm_srli_si128(uu,4); |
||
| 69 | v2 = _mm_unpacklo_epi8(v2,v2); |
||
| 70 | |||
| 71 | _mm_store_si128((__m128i*)(dst1 + 2*i), _mm_or_si128( _mm_and_si128(e0,v2), _mm_andnot_si128(e0,v1) ) ); |
||
| 72 | |||
| 73 | e0 = _mm_cmpeq_epi8(md,ll); |
||
| 74 | e0 = _mm_andnot_si128(maskall, e0); |
||
| 75 | e0 = _mm_srli_si128(e0,4); |
||
| 76 | e0 = _mm_unpacklo_epi8(e0, _mm_setzero_si128()); |
||
| 77 | |||
| 78 | e1 = _mm_cmpeq_epi8(mf,ll); |
||
| 79 | e1 = _mm_andnot_si128(maskall, e1); |
||
| 80 | e1 = _mm_srli_si128(e1,4); |
||
| 81 | e1 = _mm_unpacklo_epi8(_mm_setzero_si128(), e1); |
||
| 82 | |||
| 83 | e0 = _mm_or_si128(e0, e1); |
||
| 84 | |||
| 85 | v3 = _mm_srli_si128(ll,4); |
||
| 86 | v3 = _mm_unpacklo_epi8(v3,v3); |
||
| 87 | |||
| 88 | _mm_store_si128((__m128i*)(dst2 + 2*i), _mm_or_si128( _mm_and_si128(e0,v3), _mm_andnot_si128(e0,v1) ) ); |
||
| 89 | |||
| 90 | } else { |
||
| 91 | |||
| 92 | __m64 v0 = *(__m64*)(m+i); |
||
| 93 | __m128i v1 = _mm_movpi64_epi64(v0); |
||
| 94 | v1 = _mm_unpacklo_epi8(v1,v1); |
||
| 95 | _mm_store_si128((__m128i*)(dst1 + 2*i), v1); |
||
| 96 | _mm_store_si128((__m128i*)(dst2 + 2*i), v1); |
||
| 97 | } |
||
| 98 | } |
||
| 99 | } |
||
| 100 | |||
| 101 | #else // MMX vectorized |
||
| 102 | |||
| 103 | void lines_scale2(const unsigned char *src, unsigned y, unsigned char *dst1, unsigned char *dst2, unsigned nPix) |
||
| 104 | { |
||
| 105 | const unsigned char |
||
| 106 | *u = src + ((y-1) & 7)*sc2lines_width, |
||
| 107 | *m = src + ((y+0) & 7)*sc2lines_width, |
||
| 108 | *l = src + ((y+1) & 7)*sc2lines_width; |
||
| 109 | |||
| 110 | for (unsigned i = 0; i < nPix; i += 4) { |
||
| 111 | |||
| 112 | if (*(unsigned*)(u+i) ^ *(unsigned*)(l+i)) { |
||
| 113 | |||
| 114 | __m64 mm = *(__m64*)(m+i-2); |
||
| 115 | __m64 uu = *(__m64*)(u+i-2); |
||
| 116 | __m64 ll = *(__m64*)(l+i-2); |
||
| 117 | __m64 md = _mm_slli_si64(mm,8); |
||
| 118 | __m64 mf = _mm_srli_si64(mm,8); |
||
| 119 | __m64 maskall = _mm_or_si64(_mm_cmpeq_pi8(md,mf), _mm_cmpeq_pi8(uu,ll)); |
||
| 120 | |||
| 121 | __m64 e0, e1, v1, v2; |
||
| 122 | |||
| 123 | e0 = _mm_cmpeq_pi8(md,uu); |
||
| 124 | e0 = _mm_andnot_si64(maskall, e0); |
||
| 125 | e0 = _mm_srli_si64(e0,16); |
||
| 126 | e0 = _mm_unpacklo_pi8(e0, _mm_setzero_si64()); |
||
| 127 | |||
| 128 | e1 = _mm_cmpeq_pi8(mf,uu); |
||
| 129 | e1 = _mm_andnot_si64(maskall, e1); |
||
| 130 | e1 = _mm_srli_si64(e1,16); |
||
| 131 | e1 = _mm_unpacklo_pi8(_mm_setzero_si64(), e1); |
||
| 132 | |||
| 133 | e0 = _mm_or_si64(e0, e1); |
||
| 134 | |||
| 135 | v1 = _m_from_int(*(unsigned*)(m+i)); |
||
| 136 | v2 = _m_from_int(*(unsigned*)(u+i)); |
||
| 137 | v1 = _mm_unpacklo_pi8(v1,v1); |
||
| 138 | v2 = _mm_unpacklo_pi8(v2,v2); |
||
| 139 | |||
| 140 | *(__m64*)(dst1 + 2*i) = _mm_or_si64( _mm_and_si64(e0,v2), _mm_andnot_si64(e0,v1) ); |
||
| 141 | |||
| 142 | e0 = _mm_cmpeq_pi8(md,ll); |
||
| 143 | e0 = _mm_andnot_si64(maskall, e0); |
||
| 144 | e0 = _mm_srli_si64(e0,16); |
||
| 145 | e0 = _mm_unpacklo_pi8(e0, _mm_setzero_si64()); |
||
| 146 | |||
| 147 | e1 = _mm_cmpeq_pi8(mf,ll); |
||
| 148 | e1 = _mm_andnot_si64(maskall, e1); |
||
| 149 | e1 = _mm_srli_si64(e1,16); |
||
| 150 | e1 = _mm_unpacklo_pi8(_mm_setzero_si64(), e1); |
||
| 151 | |||
| 152 | e0 = _mm_or_si64(e0, e1); |
||
| 153 | |||
| 154 | v1 = _m_from_int(*(unsigned*)(m+i)); |
||
| 155 | v2 = _m_from_int(*(unsigned*)(l+i)); |
||
| 156 | v1 = _mm_unpacklo_pi8(v1,v1); |
||
| 157 | v2 = _mm_unpacklo_pi8(v2,v2); |
||
| 158 | |||
| 159 | *(__m64*)(dst2 + 2*i) = _mm_or_si64( _mm_and_si64(e0,v2), _mm_andnot_si64(e0,v1) ); |
||
| 160 | |||
| 161 | } else { |
||
| 162 | |||
| 163 | __m64 v1 = _m_from_int(*(unsigned*)(m+i)); |
||
| 164 | v1 = _mm_unpacklo_pi8(v1,v1); |
||
| 165 | *(__m64*)(dst1 + 2*i) = v1; |
||
| 166 | *(__m64*)(dst2 + 2*i) = v1; |
||
| 167 | |||
| 168 | } |
||
| 169 | |||
| 170 | } |
||
| 171 | } |
||
| 172 | |||
| 173 | #endif // SSE2 |
||
| 174 | |||
| 175 | #else // MMX branched |
||
| 176 | // src dst |
||
| 177 | // ABC e0e1 |
||
| 178 | // DEF e2e3 |
||
| 179 | // GHI |
||
| 180 | |||
| 181 | /* |
||
| 182 | if(B != H && D != F) |
||
| 183 | { E0 = E; |
||
| 184 | E0 = D == B ? D : E; E1 = E; |
||
| 185 | E1 = B == F ? F : E; E2 = E; |
||
| 186 | E2 = D == H ? D : E; E3 = E; |
||
| 187 | E3 = H == F ? F : E; if(B != H) continue; |
||
| 188 | } => if(D != F) |
||
| 189 | else { |
||
| 190 | { E0 = D == B ? D : E; |
||
| 191 | E0 = E; E1 = B == F ? F : E; |
||
| 192 | E1 = E; E2 = D == H ? D : E; |
||
| 193 | E2 = E; E3 = H == F ? F : E; |
||
| 194 | E3 = E; } |
||
| 195 | } |
||
| 196 | */ |
||
| 197 | |||
| 198 | void lines_scale2(const unsigned char *src, unsigned y, unsigned char *dst1, unsigned char *dst2, unsigned nPix) |
||
| 199 | { |
||
| 200 | const unsigned char |
||
| 201 | *u = src + ((y-1) & 7)*sc2lines_width, |
||
| 202 | *m = src + ((y+0) & 7)*sc2lines_width, |
||
| 203 | *l = src + ((y+1) & 7)*sc2lines_width; |
||
| 204 | |||
| 205 | // process 4pix per iteration |
||
| 206 | for (unsigned i = 0; i < nPix; i += 4) |
||
| 207 | { |
||
| 208 | unsigned dw = *(unsigned*)(m+i); |
||
| 209 | __m64 v1 = _mm_cvtsi32_si64(dw); // v1 = 0| 0| 0| 0|dw[3]|dw[2]|dw[1]|dw[0] |
||
| 210 | v1 = _mm_unpacklo_pi8(v1,v1); // v1 = dw[3]|dw[3]|dw[2]|dw[2]|dw[1]|dw[1]|dw[0]|dw[0] |
||
| 211 | *(__m64*)(dst1 + 2*i) = v1; // e0e1 = dw[3]|dw[3]|dw[2]|dw[2]|dw[1]|dw[1]|dw[0]|dw[0] |
||
| 212 | *(__m64*)(dst2 + 2*i) = v1; // e2e3 = dw[3]|dw[3]|dw[2]|dw[2]|dw[1]|dw[1]|dw[0]|dw[0] |
||
| 213 | |||
| 214 | dw = *(unsigned*)(u+i) ^ *(unsigned*)(l+i); |
||
| 215 | if (!dw) |
||
| 216 | continue; // u == l |
||
| 217 | |||
| 218 | #define process_pix(n) \ |
||
| 219 | if ((dw & (0xFF << (8*n))) && m[i+n-1] != m[i+n+1]) \ |
||
| 220 | { \ |
||
| 221 | if (u[i+n] == m[i+n-1]) \ |
||
| 222 | dst1[2*(i+n)] = u[i+n]; \ |
||
| 223 | if (u[i+n] == m[i+n+1]) \ |
||
| 224 | dst1[2*(i+n)+1] = u[i+n]; \ |
||
| 225 | if (l[i+n] == m[i+n-1]) \ |
||
| 226 | dst2[2*(i+n)] = l[i+n]; \ |
||
| 227 | if (l[i+n] == m[i+n+1]) \ |
||
| 228 | dst2[2*(i+n)+1] = l[i+n]; \ |
||
| 229 | } |
||
| 230 | |||
| 231 | process_pix(0); |
||
| 232 | process_pix(1); |
||
| 233 | process_pix(2); |
||
| 234 | process_pix(3); |
||
| 235 | #undef process_pix |
||
| 236 | } |
||
| 237 | } |
||
| 238 | |||
| 239 | #endif // MMX branched |
||
| 240 | |||
| 241 | void lines_scale2_32(const unsigned char *src, unsigned y, unsigned char *dst1, unsigned char *dst2, unsigned nPix) |
||
| 242 | { |
||
| 243 | const u32 *s = (u32 *)src; |
||
| 244 | const u32 *u = s + ((y-1) & 7)*sc2lines_width; |
||
| 245 | const u32 *m = s + ((y+0) & 7)*sc2lines_width; |
||
| 246 | const u32 *l = s + ((y+1) & 7)*sc2lines_width; |
||
| 247 | u32 *d1 = (u32 *)dst1; |
||
| 248 | u32 *d2 = (u32 *)dst2; |
||
| 249 | |||
| 250 | for (unsigned i = 0; i < nPix; i++) |
||
| 251 | { |
||
| 252 | d1[2*i] = d1[2*i+1] = d2[2*i] = d2[2*i+1] = m[i]; |
||
| 253 | |||
| 254 | if (u[i] != l[i] && m[i-1] != m[i+1]) |
||
| 255 | { |
||
| 256 | if (u[i] == m[i-1]) |
||
| 257 | d1[2*i] = u[i]; |
||
| 258 | if (u[i] == m[i+1]) |
||
| 259 | d1[2*i+1] = u[i]; |
||
| 260 | if (l[i] == m[i-1]) |
||
| 261 | d2[2*i] = l[i]; |
||
| 262 | if (l[i] == m[i+1]) |
||
| 263 | d2[2*i+1] = l[i]; |
||
| 264 | } |
||
| 265 | } |
||
| 266 | } |
||
| 267 | |||
| 268 | // 8bpp |
||
| 269 | void render_scale2(unsigned char *dst, unsigned pitch) |
||
| 270 | { |
||
| 271 | unsigned char *src = rbuf; unsigned delta = temp.scx/4; |
||
| 272 | line_8_any(t.scale2buf[0], src); |
||
| 273 | // assume 'above' screen line same as line 0 |
||
| 274 | memcpy(t.scale2buf[(0-1) & 7], t.scale2buf[0], temp.scx); |
||
| 275 | for (unsigned y = 0; y < temp.scy; y++) |
||
| 276 | { |
||
| 277 | src += delta; |
||
| 278 | line_8_any(t.scale2buf[(y+1) & 7], src); |
||
| 279 | lines_scale2(t.scale2buf[0], y, dst, dst+pitch, temp.scx); |
||
| 280 | dst += 2*pitch; |
||
| 281 | } |
||
| 282 | } |
||
| 283 | |||
| 284 | // 32bpp |
||
| 285 | void render_scale2_32(unsigned char *dst, unsigned pitch) |
||
| 286 | { |
||
| 287 | unsigned char *src = rbuf; |
||
| 288 | unsigned delta = temp.scx/4; |
||
| 289 | line_32_any((u8 *)t.scale2buf32[0], src); |
||
| 290 | |||
| 291 | // assume 'above' screen line same as line 0 |
||
| 292 | memcpy(t.scale2buf32[(0-1) & 7], t.scale2buf32[0], temp.scx); |
||
| 293 | for (unsigned y = 0; y < temp.scy; y++) |
||
| 294 | { |
||
| 295 | src += delta; |
||
| 296 | line_32_any((u8 *)t.scale2buf32[(y+1) & 7], src); |
||
| 297 | lines_scale2_32((u8 *)t.scale2buf32[0], y, dst, dst+pitch, temp.scx); |
||
| 298 | dst += 2*pitch; |
||
| 299 | } |
||
| 300 | } |
||
| 301 | |||
| 302 | // MMX-vectorized version is not ready yet :( |
||
| 303 | // 8bpp |
||
| 304 | void lines_scale3(unsigned y, unsigned char *dst, unsigned pitch) |
||
| 305 | { |
||
| 306 | |||
| 307 | const unsigned char |
||
| 308 | *u = t.scale2buf[(y-1) & 3], |
||
| 309 | *m = t.scale2buf[(y+0) & 3], |
||
| 310 | *l = t.scale2buf[(y+1) & 3]; |
||
| 311 | |||
| 312 | for (unsigned i = 0; i < temp.scx; i += 4) |
||
| 313 | { |
||
| 314 | unsigned char c; |
||
| 315 | |||
| 316 | c = m[i]; |
||
| 317 | dst[3*i+0+0*pitch+ 0] = dst[3*i+1+0*pitch+ 0] = dst[3*i+2+0*pitch+ 0] = c; |
||
| 318 | dst[3*i+0+1*pitch+ 0] = dst[3*i+1+1*pitch+ 0] = dst[3*i+2+1*pitch+ 0] = c; |
||
| 319 | dst[3*i+0+2*pitch+ 0] = dst[3*i+1+2*pitch+ 0] = dst[3*i+2+2*pitch+ 0] = c; |
||
| 320 | |||
| 321 | c = m[i+1]; |
||
| 322 | dst[3*i+0+0*pitch+ 3] = dst[3*i+1+0*pitch+ 3] = dst[3*i+2+0*pitch+ 3] = c; |
||
| 323 | dst[3*i+0+1*pitch+ 3] = dst[3*i+1+1*pitch+ 3] = dst[3*i+2+1*pitch+ 3] = c; |
||
| 324 | dst[3*i+0+2*pitch+ 3] = dst[3*i+1+2*pitch+ 3] = dst[3*i+2+2*pitch+ 3] = c; |
||
| 325 | |||
| 326 | c = m[i+2]; |
||
| 327 | dst[3*i+0+0*pitch+ 6] = dst[3*i+1+0*pitch+ 6] = dst[3*i+2+0*pitch+ 6] = c; |
||
| 328 | dst[3*i+0+1*pitch+ 6] = dst[3*i+1+1*pitch+ 6] = dst[3*i+2+1*pitch+ 6] = c; |
||
| 329 | dst[3*i+0+2*pitch+ 6] = dst[3*i+1+2*pitch+ 6] = dst[3*i+2+2*pitch+ 6] = c; |
||
| 330 | |||
| 331 | c = m[i+3]; |
||
| 332 | dst[3*i+0+0*pitch+ 9] = dst[3*i+1+0*pitch+ 9] = dst[3*i+2+0*pitch+ 9] = c; |
||
| 333 | dst[3*i+0+1*pitch+ 9] = dst[3*i+1+1*pitch+ 9] = dst[3*i+2+1*pitch+ 9] = c; |
||
| 334 | dst[3*i+0+2*pitch+ 9] = dst[3*i+1+2*pitch+ 9] = dst[3*i+2+2*pitch+ 9] = c; |
||
| 335 | |||
| 336 | unsigned dw = *(unsigned*)(u+i) ^ *(unsigned*)(l+i); |
||
| 337 | if (!dw) continue; |
||
| 338 | |||
| 339 | #define process_pix(n) \ |
||
| 340 | if ((dw & (0xFF << (8*n))) && m[i+n-1] != m[i+n+1]) \ |
||
| 341 | { \ |
||
| 342 | if (u[i+n] == m[i+n-1]) \ |
||
| 343 | dst[0*pitch+3*(i+n)] = u[i+n]; \ |
||
| 344 | if ((u[i+n] == m[i+n-1] && m[i+n] != u[i+n+1]) || (u[i+n] == m[i+n+1] && m[i+n] != u[i+n-1])) \ |
||
| 345 | dst[0*pitch+3*(i+n)+1] = u[i+n]; \ |
||
| 346 | if (u[i+n] == m[i+n+1]) \ |
||
| 347 | dst[0*pitch+3*(i+n)+2] = u[i+n]; \ |
||
| 348 | if ((u[i+n] == m[i+n-1] && m[i+n] != l[i+n-1]) || (l[i+n] == m[i+n-1] && m[i+n] != u[i+n-1])) \ |
||
| 349 | dst[1*pitch+3*(i+n)+0] = m[i+n-1]; \ |
||
| 350 | if ((u[i+n] == m[i+n+1] && m[i+n] != l[i+n+1]) || (l[i+n] == m[i+n+1] && m[i+n] != u[i+n+1])) \ |
||
| 351 | dst[1*pitch+3*(i+n)+2] = m[i+n+1]; \ |
||
| 352 | if (l[i+n] == m[i+n-1]) \ |
||
| 353 | dst[2*pitch+3*(i+n)] = l[i+n]; \ |
||
| 354 | if ((l[i+n] == m[i+n-1] && m[i+n] != l[i+n+1]) || (l[i+n] == m[i+n+1] && m[i+n] != l[i+n-1])) \ |
||
| 355 | dst[2*pitch+3*(i+n)+1] = l[i+n]; \ |
||
| 356 | if (l[i+n] == m[i+n+1]) \ |
||
| 357 | dst[2*pitch+3*(i+n)+2] = l[i+n]; \ |
||
| 358 | } |
||
| 359 | |||
| 360 | process_pix(0); |
||
| 361 | process_pix(1); |
||
| 362 | process_pix(2); |
||
| 363 | process_pix(3); |
||
| 364 | #undef process_pix |
||
| 365 | } |
||
| 366 | } |
||
| 367 | |||
| 368 | // 8bpp |
||
| 369 | void render_scale3(unsigned char *dst, unsigned pitch) |
||
| 370 | { |
||
| 371 | unsigned char *src = rbuf; unsigned delta = temp.scx/4; |
||
| 372 | line_8_any(t.scale2buf[0], src); |
||
| 373 | // assume 'above' screen line same as line 0 |
||
| 374 | memcpy(t.scale2buf[(0-1) & 3], t.scale2buf[0], temp.scx); |
||
| 375 | for (unsigned y = 0; y < temp.scy; y++) { |
||
| 376 | src += delta; |
||
| 377 | line_8_any(t.scale2buf[(y+1) & 3], src); |
||
| 378 | lines_scale3(y, dst, pitch); |
||
| 379 | dst += 3*pitch; |
||
| 380 | } |
||
| 381 | } |
||
| 382 | |||
| 383 | // 32bpp |
||
| 384 | void lines_scale3_32(unsigned y, unsigned char *dst, unsigned pitch) |
||
| 385 | { |
||
| 386 | const u32 *u = t.scale2buf32[(y-1) & 3]; |
||
| 387 | const u32 *m = t.scale2buf32[(y+0) & 3]; |
||
| 388 | const u32 *l = t.scale2buf32[(y+1) & 3]; |
||
| 389 | u32 *d = (u32 *)dst; |
||
| 390 | pitch /= sizeof(u32); |
||
| 391 | |||
| 392 | for (unsigned i = 0; i < temp.scx; i++) |
||
| 393 | { |
||
| 394 | d[0*pitch+3*i+0] = d[0*pitch+3*i+1] = d[0*pitch+3*i+2] = m[i]; |
||
| 395 | d[1*pitch+3*i+0] = d[1*pitch+3*i+1] = d[1*pitch+3*i+2] = m[i]; |
||
| 396 | d[2*pitch+3*i+0] = d[2*pitch+3*i+1] = d[2*pitch+3*i+2] = m[i]; |
||
| 397 | |||
| 398 | if (u[i] != l[i] && m[i-1] != m[i+1]) |
||
| 399 | { |
||
| 400 | if (u[i] == m[i-1]) |
||
| 401 | d[0*pitch+3*i+0] = u[i]; |
||
| 402 | if ((u[i] == m[i-1] && m[i] != u[i+1]) || (u[i] == m[i+1] && m[i] != u[i-1])) |
||
| 403 | d[0*pitch+3*i+1] = u[i]; |
||
| 404 | if (u[i] == m[i+1]) |
||
| 405 | d[0*pitch+3*i+2] = u[i]; |
||
| 406 | if ((u[i] == m[i-1] && m[i] != l[i-1]) || (l[i] == m[i-1] && m[i] != u[i-1])) |
||
| 407 | d[1*pitch+3*i+0] = m[i-1]; |
||
| 408 | if ((u[i] == m[i+1] && m[i] != l[i+1]) || (l[i] == m[i+1] && m[i] != u[i+1])) |
||
| 409 | d[1*pitch+3*i+2] = m[i+1]; |
||
| 410 | if (l[i] == m[i-1]) |
||
| 411 | d[2*pitch+3*i+0] = l[i]; |
||
| 412 | if ((l[i] == m[i-1] && m[i] != l[i+1]) || (l[i] == m[i+1] && m[i] != l[i-1])) |
||
| 413 | d[2*pitch+3*i+1] = l[i]; |
||
| 414 | if (l[i] == m[i+1]) |
||
| 415 | d[2*pitch+3*i+2] = l[i]; |
||
| 416 | } |
||
| 417 | } |
||
| 418 | } |
||
| 419 | |||
| 420 | // 32bpp |
||
| 421 | void render_scale3_32(unsigned char *dst, unsigned pitch) |
||
| 422 | { |
||
| 423 | unsigned char *src = rbuf; unsigned delta = temp.scx/4; |
||
| 424 | line_32_any((u8 *)t.scale2buf32[0], src); |
||
| 425 | // assume 'above' screen line same as line 0 |
||
| 426 | memcpy(t.scale2buf32[(0-1) & 3], t.scale2buf32[0], temp.scx); |
||
| 427 | for (unsigned y = 0; y < temp.scy; y++) |
||
| 428 | { |
||
| 429 | src += delta; |
||
| 430 | line_32_any((u8 *)t.scale2buf32[(y+1) & 3], src); |
||
| 431 | lines_scale3_32(y, dst, pitch); |
||
| 432 | dst += 3*pitch; |
||
| 433 | } |
||
| 434 | } |
||
| 435 | |||
| 436 | void render_scale4(unsigned char *dst, unsigned pitch) |
||
| 437 | { |
||
| 438 | unsigned char *src = rbuf; unsigned delta = temp.scx/4; |
||
| 439 | |||
| 440 | line_8_any(t.scale2buf[0], src); src += delta; |
||
| 441 | line_8_any(t.scale2buf[1], src); src += delta; |
||
| 442 | // assume 'above' screen line same as line 0 |
||
| 443 | memcpy(t.scale2buf[(0-1) & 7], t.scale2buf[0], temp.scx); |
||
| 444 | lines_scale2(t.scale2buf[0], 0, t.scale4buf[0], t.scale4buf[1], temp.scx); |
||
| 445 | |||
| 446 | for (unsigned y = 0; y < temp.scy; y++) { |
||
| 447 | |||
| 448 | line_8_any(t.scale2buf[(y+2) & 7], src); src += delta; |
||
| 449 | |||
| 450 | unsigned char *dst1 = t.scale4buf[(2*y+2) & 7]; |
||
| 451 | unsigned char *dst2 = t.scale4buf[(2*y+3) & 7]; |
||
| 452 | lines_scale2(t.scale2buf[0], y+1, dst1, dst2, temp.scx); |
||
| 453 | |||
| 454 | lines_scale2(t.scale4buf[0], 2*y, dst+0*pitch, dst+1*pitch, temp.scx*2); |
||
| 455 | lines_scale2(t.scale4buf[0], 2*y+1, dst+2*pitch, dst+3*pitch, temp.scx*2); |
||
| 456 | |||
| 457 | dst += 4*pitch; |
||
| 458 | } |
||
| 459 | } |
||
| 460 | |||
| 461 | void __fastcall render_advmame(unsigned char *dst, unsigned pitch) |
||
| 462 | { |
||
| 463 | switch (conf.videoscale) |
||
| 464 | { |
||
| 465 | case 2: |
||
| 466 | if(temp.obpp == 8) render_scale2(dst, pitch); |
||
| 467 | else if(temp.obpp == 32) render_scale2_32(dst, pitch); |
||
| 468 | break; |
||
| 469 | case 3: |
||
| 470 | if(temp.obpp == 8) render_scale3(dst, pitch); |
||
| 471 | else if(temp.obpp == 32) render_scale3_32(dst, pitch); |
||
| 472 | break; |
||
| 473 | case 4: render_scale4(dst, pitch); break; |
||
| 474 | default: render_small(dst, pitch); return; // skip noflic test |
||
| 475 | } |
||
| 476 | if (conf.noflic) |
||
| 477 | memcpy(rbuf_s, rbuf, temp.scy*temp.scx/4); |
||
| 478 | _mm_empty(); |
||
| 479 | } |